Mapping and spatial data

import polars as pl
from pyobsplot import Obsplot, Plot, js

The Geo mark allows to draw geographic features such as points, lines and polygons. These marks data are passed as GeoJSON.

This allows to create choropleth maps such as the following:

import json

# Load US counties from GeoJson
# Source : https://eric.clst.org/tech/usgeojson/
with open("data/us_counties.geojson", "r", encoding="windows-1252") as f:
    counties = json.load(f)

# Merge unemployment values as GeoJson features property
unemployment = pl.read_csv("data/us-county-unemployment.csv")
rates = {
    id: unemp
    for id, unemp in zip(unemployment.get_column("id"), unemployment.get_column("rate"))
}
for county in counties["features"]:
    key = int(county["properties"]["GEO_ID"][-5:])
    if key in rates:
        county["properties"]["unemployment"] = rates[key]

# Map
Obsplot(
    {
        "marks": [Plot.geo(counties, {"fill": js("(d) => d.properties.unemployment")})],
        # "marks": [Plot.geo(counties, {"stroke": "red"})],
        "projection": "albers-usa",
        "color": {
            "type": "quantile",
            "n": 8,
            "scheme": "blues",
            "label": "Unemployment (%)",
            "legend": True,
        },
    }
)

Geographic data in Observable notebooks are often stored in TopoJSON files. The topojson package can be useful for these kind of data.

import topojson as tp

# Import TopoJSON data
with open("data/us-counties-10m.json", "r") as f:
    counties = json.load(f)

# Extract "nation" objects and convert to GeoJSON
nation = tp.Topology(counties, object_name="nation").to_geojson(object_name="nation")
nation = json.loads(nation)
# Extract "states" objects and convert to GeoJSON
states = tp.Topology(counties, object_name="states").to_geojson(object_name="states")
states = json.loads(states)

# Plot map
Obsplot(
    {
        "marks": [
            Plot.geo(states, {"strokeOpacity": 0.3}),
            Plot.geo(nation),
        ],
        "projection": "albers",
    }
)

Of course other marks can be used in conjunction with geo marks. This example represents the density of Walmarts supermarkets and is taken from the Mapping notebook.

walmarts = pl.read_csv("data/walmarts.tsv", sep="\t")

Obsplot(
    {
        "marks": [
            Plot.density(
                walmarts,
                {"x": "longitude", "y": "latitude", "bandwidth": 12, "fill": "density"},
            ),
            Plot.dot(
                walmarts,
                {"x": "longitude", "y": "latitude", "r": 1, "fill": "currentColor"},
            ),
            Plot.geo(states, {"strokeOpacity": 0.3, "fill": None}),
            Plot.geo(nation),
        ],
        "projection": "albers",
        "color": {"scheme": "blues"},
    }
)

/tmp/ipykernel_2369/2759925357.py:1: DeprecationWarning: `sep` is deprecated as an argument to `read_csv`; use `separator` instead.
  walmarts = pl.read_csv("data/walmarts.tsv", sep="\t")

The Raster mark creates an image from spatial data.

Plot allows to do different type of spatial interpolations, such as nearest, which draws voronoi cells around values:

ca55 = pl.read_csv("data/ca55-south.csv")


def flare_map(interpolation):
    return Obsplot(
        {
            "x": {"axis": None},
            "y": {"axis": None},
            "inset": 10,
            "marginBottom": 2,
            "height": 500,
            "color": {"type": "diverging"},
            "marks": [
                Plot.raster(
                    ca55,
                    {
                        "x": "LONGITUDE",
                        "y": "LATITUDE",
                        "fill": "MAG_IGRF90",
                        "interpolate": interpolation,
                    },
                ),
                Plot.frame(),
            ],
        }
    )


flare_map("nearest")

Or the more recent random walk interpolation:

flare_map("random-walk")